Tree Genetics & Genomes (2019) 15:18 https://doi.org/10.1007/s11295-019-1324-y

ORIGINAL ARTICLE

Two large-effect QTLs, Ma and Ma3, determine genetic potential for acidity in fruit: breeding insights from a multi-family study

S. Verma1,2 & K. Evans3 & Y. Guan3 & J. J. Luby4 & U. R. Rosyara5,6 & N. P. Howard 4,7 & N. Bassil8 & M. C. A. M. Bink9,10 & W. E. van de Weg11 & C. P. Peace1

Received: 23 September 2018 /Revised: 23 January 2019 /Accepted: 28 January 2019 # The Author(s) 2019

Abstract Acidity is a critical component of the apple fruit consumption experience. In previous biparental family studies, two large-effect acidity QTLs were reported using freshly harvested fruit. Objectives of this study were to determine the number and location of QTLs for acidity variation in a large apple breeding program and ascertain the quantitative effects and breeding relevance of QTL allelic combinations at harvest and after commercially relevant periods of cold storage. Pedigree-connected germplasm of 16 full-sib families representing nine important breeding parents, genotyped for the 8K SNP array, was assessed for titratable acidity at harvest and after 10- and 20-week storage treatments, for three successive seasons. Using pedigree-based QTL mapping software, FlexQTL™, evidence was found for only two QTLs, on linkage groups 16 (the reported Ma locus) and LG 8 (here called Ma3) that jointly explained 66 ± 5% of the phenotypic variation. An additive allele dosage model for the two QTLs effectively explained most acidity variation, with an average of + 1.8 mg/L at harvest per high-acidity allele. The more high-acidity alleles, the faster the depletion with storage, with all combinations appearing to eventually converge to a common baseline. All parent and selections had one or two of the four possible high-acidity alleles. Each QTL had a rare second high-acidity allele with stronger or reduced effect. Diagnostic SNP markers were identified for QTL alleles derived from distinct sources. Combined QTL effects highlighted utility of the DNA- based information in new development for targeting desired fruit acidity levels before or after storage.

Keywords FlexQTL™ . Malic acid . × domestica . Pedigree-Based Analysis . RosBREED

Communicated by D. Chagné Electronic supplementary material The online version of this article (https://doi.org/10.1007/s11295-019-1324-y) contains supplementary material, which is available to authorized users.

* C. P. Peace 6 Present address: CIMMYT International, Global Wheat Program, CP [email protected] 56237 Texcoco, Mexico

7 1 Department of Horticulture, Washington State University, Present address: Institut für Biologie und Umweltwissenschaften, Pullman, WA 99164, USA Carl von Ossietzky Universität, 26129 Oldenburg, Germany 2 Present address: Department of Horticultural Science, IFAS Gulf 8 National Clonal Germplasm Repository, Agricultural Research Research and Education Center, University of Florida, Service, United States Department of Agriculture, Wimauma, FL 33598, USA Corvallis, OR 97333, USA 3 Tree Fruit Research and Extension Center, Department of 9 Biometris, Wageningen UR, PO Box 16, 6700 Horticulture, Washington State University, Wenatchee, WA 98801, AA Wageningen, The Netherlands USA 10 4 Department of Horticultural Science, University of Minnesota, St. Present address: Research & Technology Center, Hendrix Genetics, Paul, MN 55108, USA 5830 AC Boxmeer, The Netherlands 5 Department of Horticulture, Michigan State University, East 11 Plant Breeding, Wageningen UR, 6708 Lansing, MI 48824, USA PB Wageningen, The Netherlands 18 Page 2 of 17 Tree Genetics & Genomes (2019) 15:18

Introduction described a second significant QTL for acidity on LG 8 that explained 46% of the phenotypic variation in a cross between Acidity is a critical component of the apple fruit consumption ‘’ and ‘’ that also segregated for Ma (42% of experience. Apple fruit are popular worldwide, consumed as the phenotypic variation). Similarly, Kenis et al. (2008)using either fresh or processed products and contributing to human the biparental family ‘Telamon’ × ‘’ reported signif- health and well-being (Iwanami et al. 2012). Fruit acidity af- icant QTLs for acidity at both loci: in the vicinity of the Ma fects overall apple flavor and the perception of other flavor locus in two consecutive years as well as on LG 8, although traits such as sweetness and aroma, therefore influencing con- the latter was detected in only one of the years. Zhang et al. sumer satisfaction (Harker et al. 2003; Iwanami et al. 2012). (2012), using the biparental family ‘’ × ‘Golden Malic acid is the organic acid in mature apple fruit Delicious,’ reported a LG 8 QTL for malic acid that explained (Zhang et al. 2010), and can be measured sensorially by taste 13.5% of the phenotypic variation (total organic acid content or instrumentally by titration or as organic acid content of fruit assessed by gas chromatography). Ma et al. (2016)reported juice. Commercially, most apple fruit are stored then eaten QTL at both loci that explained 12% and 14% of the pheno- months to a year after harvest because are harvested typic variation. Kumar et al. (2012)alsoreportedQTLsfor for only 3 to 4 months yet available to consumers all year acidity on LG 8 and LG 16 in a genomic selection study with a round (Watkins 2017). During storage, malic acid content de- training population of seven full-sib families. King et al. clines (Haynes 1925; Kouassi et al. 2009) and can reach low (2000), using the same mapping population as Maliepaard levels associated with lower satisfaction for some segments of et al. (1998), reported a QTL for acidity in the vicinity of the consumers (Harker et al. 2008). However, there is genetic Ma locus, with significant influences also on firmness and sen- variation for acidity levels before and after storage among sory traits of crispness, juiciness, sponginess, and overall liking. common cultivars (Rouchaud et al. 1985;Iwanamietal. Minor QTLs for acidity were reported on LGs 2, 10, 13, 15, and 2008) and among germplasm in breeding programs (Ma 17 in the ‘Telamon’ × ‘Braeburn’ family (Kenis et al. 2008). In et al. 2015b; Cliff et al. 2016). An understanding of the com- acrossbetween‘Royal ’ and the accession ponents of that genetic variation for fruit acidity would aid PI 613988, Xu et al. (2012) detected a major QTL at the Ma development of new apple cultivars that produce fruit with locus and minor QTLs for acidity on LGs 1 and 6. target levels of acidity. Allelic variation in genes of the biochemical pathways un- Genetic studies of apple fruit acidity have revealed the derlying fruit acidity biosynthesis and degradation or in genes presence of major genetic factors. Wellington (1924)de- that affect the transport, storage, and use of acids might un- scribed Bsweet flavor^ (essentially the absence of strong acid- derlie the detected QTLs. Several biochemical pathways are ity) as a recessive character. By evaluating apple seedlings on involved in the synthesis and metabolism of malate, the pre- a five-point sensory scale from sour to sweet, Crane and cursor of malic acid. In the glycolysis pathway, carboxylation Lawrence (1934) ascribed homozygosity or heterozygosity of phosphoenol pyruvate occurs in the cytoplasm, producing to more than 20 apple cultivars and almost 600 seedlings. oxaloacetate, which is reduced to the volatile compound ma- Nybom (1959) and Visser and Verhaegh (1978) using inde- late by NAD-dependent malate dehydrogenase (Etienne et al. pendent segregation ratio analysis and analysis of variances, 2013). Other pathways that could also influence cytoplasmic respectively, confirmed the observation that low acidity malate content are as follows: the conversion of malate from behaves as a recessive phenotype although it is rarely pyruvate by NADP-linked malic enzyme in the chloroplast; observed among apple cultivars and selections. Dove and the conversion to malate of glyoxylate and oxaloacetic acid by Murphy (1936) and Nybom (1959) suggested that early seed- malate synthase and NAD-linked malate dehydrogenase, re- ling selection for fruit acidity levels could be conducted using spectively, in the glyoxysome; the conversion of malate from the indirect measure of leaf pH. Visser and Verhaegh (1978) fumarate by fumarase in mitochondria; and the efflux of ma- introduced the Malic acid (Ma) locus to explain these obser- late from the vacuole into the cytoplasm (Sweetman et al. vations of relatively simple inheritance. 2009). Once fruit are harvested, continued respiration depletes Quantitative trait loci (QTLs) for apple fruit acidity have available cellular malate, underlying the typical reduction in been located in the apple genome, especially at the Ma locus. acidity throughout the postharvest storage period (Etienne Maliepaard et al. (1998) were the first to map the Ma locus as a et al. 2013). Bai et al. (2012) provided strong evidence that morphological marker to the top of linkage group (LG) 16 underlying the Ma locus is an aluminum-activated malate using a biparental F1 mapping family for which both parents transporter gene (ALMT1) coding for the protein responsible were heterozygous. Using QTL interval mapping with com- for transporting malic acid molecules from the cytosol to the mon cultivars as parents, numerous studies reported incom- vacuole where ~ 90% of cell sap is stored. The transport of plete dominance of the high-acidity BMa^ allele over the low- malate over the vacuole is dependent on the acidity level of the acid Bma^ allele (King et al. 2000, 2001; Liebhard et al. 2003; vacuole and membrane potential across the tonoplast Xu et al. 2012;Khanetal.2013). Liebhard et al. (2003) (Martinoia et al. 2007). More acidic conditions within the Tree Genetics & Genomes (2019) 15:18 Page 3 of 17 18 vacuole can lead to a change in the malate’s protonation state, The objectives of this study were to determine the number which can further support accumulation, especially if the and location of QTLs for acidity variation in a large apple transporter involved is specific for one form of malate breeding program and ascertain the quantitative effects and (Martinoia et al. 2007). In plants, the proton pumps V- breeding relevance of QTL allelic combinations at harvest ATPase and V-PPase are involved in vacuolar acidification and after commercially relevant periods of cold storage. (Martinoia et al. 2007). In apple, such genes have been iden- tified, and their involvement in vacuolar acidification and ma- late accumulation has been confirmed (Hu et al. 2016, 2017; Materials and methods Jia et al. 2018). Coordination of these proton pumps and ma- late transporters is under research and seems to involve a Breeding germplasm complex network of regulating and signaling genes that in- clude the MYB genes MdMYB1, MdMYB44,andMdMYB73 The germplasm set used was a subset of the US apple Crop (Hu et al. 2016, 2017; Jia et al. 2018)ofwhichMdMYB73 is Reference Set supplemented with a Breeding Pedigree Set influenced by, among others, the MdCIbHLH1 gene involved specific to the WSU program and chosen to efficiently repre- in cold tolerance (Hu et al. 2017). sent genome-wide alleles of important breeding parents of the Knowledge gained from genetic dissection of apple acidity WSU breeding program as described by Peace et al. (2014). is not yet sufficient for widespread exploitation in breeding. Parentage records were verified first using three SSR markers,

For the most consistently detected trait loci on LG 16 (Ma) Md-Exp7SSR (Costa et al. 2008), CH05c06, and Hi04e04 and LG 8 (hereby termed Ma3), several unknowns remain. It (Silfverberg-Dilworth et al. 2006; Velasco et al. 2010) and is unclear whether the QTL effects are maintained across ge- then with SNP array markers using procedures described in nomic backgrounds of typical breeding germplasm. The high- Pikunova et al. (2013) and van de Weg et al. (2018). The final or low-acidity alleles reported across the various families pedigree-connected germplasm consisted of 16 F1 full-sib might be identical-by-descent or represent distinct sources that families with 200 offspring (Online Resource 1)representing might differ in quantitative effects. How allelic combinations nine important breeding parents (‘Arlet,’‘Aurora Golden at the two loci interact to produce genetic potential for fruit Gala,’‘,’‘Delicious,’‘,’‘,’ acidity is unexplored. Other QTLs might also significantly ‘,’‘WA 5,’ and selection W1), 25 of their ancestors, contribute to acidity genetic potential in wider germplasm and 18 genetically related cultivars and selections supporting than that represented by the biparental family parents. the phasing of marker data in ancestral generations. Reports described influences of the loci on freshly harvested fruit, but the more commercially pertinent effects on fruit after Phenotypic data storage are not known. Determining the frequency of acidity- associated alleles and their distributions across breeding germ- Crop load management and fruit maturity assessment were plasm would help determine the breeding relevance of signif- conducted according to Evans et al. (2012). Titratable acidity icant loci. (TA, in mg/L malic acid equivalents) using an Auto-Titrator Key experimental resources are now in place to character- (Metrohm 815 Robotic USB Sample Processor XL, Metrohm ize the major genetic factors influencing apple fruit acidity at USA, Inc., Riverview, FL, USA) was determined for fruit of harvest and after storage for diverse breeding germplasm. the germplasm set over three harvest years with three storage Genetic improvement in fruit quality via superior new scion treatments within each year: at harvest with no storage (H); cultivars that can satisfy both consumer and industry needs is after 10 weeks of cold storage followed by 1 week at room a primary target of the multi-institutional, US-wide, and inter- temperature (10wk); and after 20 weeks of cold storage national RosBREED project (Iezzoni et al. 2010; www. followed by 1 week at room temperature (20wk), as described rosbreed.org) and the participating Washington State by Evans et al. (2012). These storage periods were chosen for University (WSU) Apple Breeding Program (Evans 2013). their commercial relevance. For each assessment, the juice of The pedigree-based analysis (PBA) approach (van de Weg equally sized portions of preferably five but at least four fruit et al. 2004;Binketal.2012, 2014) has been established for were pooled. Phenotypic data were checked for errors and the aforementioned breeding program in the RosBREED pro- statistical outliers by plotting phenotypic correlations among ject, enabling efficient detection and validation of segregating pairs of years within a storage treatment and plotting TA levels alleles in multiple families (Peace et al. 2014). The US apple over storage within each year. To identify erroneous data with- Crop Reference Set (CR Set), representing alleles of important in expected bounds for an individual but outside the overrid- parents of US apple breeding germplasm, has been established ing trend of decreasing TA levels over storage, triplets of data (Peace et al. 2014) and phenotyped for numerous traits, in- points over the three storage periods for each individual within cluding fruit quality after storage, using standardized proto- each year were examined for TA increases of more than cols (Evans et al. 2012). 1.5 mg/L from H to 10wk or from 10wk to 20wk. Twenty- 18 Page 4 of 17 Tree Genetics & Genomes (2019) 15:18 one such single deviating data points were removed (Online Genotypic data Resources 2 and 3). Inspection of each individual triplet iden- tified one more individual with an unusually low acidity at Progenies and available progenitors were genotyped with the 20wk in 2012, for which a data point was then made missing International RosBREED SNP Consortium apple 8K SNP array (Online Resource 4). To identify erroneous data across years, v1, an Infinium™ array (Chagné et al. 2012a). GenomeStudio™ triplets across years for each individual at H were examined genotyping software was used for marker calling. Next, ASSIsT for cases of one data point being at least 2 mg/L higher or software (Di Guardo et al. 2015) was used for filtering of SNP lower in TA than the other two data points, and were subse- markers, whereby markers with recognized null alleles were quently made missing (Online Resource 5). To avoid biases removed, and to prepare an input file in FlexQTL™ format via from prior breeding selection for superior performance (Peace software FlexQTL DataPrepper (www.flexqtl.nl). Marker data et al. 2014), only the phenotypic data of offspring within the were curated using marker consistency, double recombination full-sib families were used for QTL analyses. The original reports, graphical genotyping plots as generated by FlexQTL™ phenotypic data set prior to the above-described curation is and VisualFlexQTL (Bink et al. 2014, www.flexqtl.nl), and provided in Online Resource 6, together with the curated data problematic markers were eliminated and double-recombinant set. singletons were made missing through FlexQTL™ parameter settings. Genetic positions of the filtered 1344 SNP markers for QTL analysis were obtained from an early version of Howard Heritability et al. (2017), with an average position difference of 0.33 cM and with any additonal SNPs (such as ss475879082, ss475882553, A linear mixed model (individual or animal model) was fitted and ss475883359) genetically positioned by relative physical to the titratable acidity data using univariate analyses in location. SNP data on ancestors that had not been genotyped ASReml software 4.1: within the RosBREED project were complemented where feasi- ble with 20K Infinium™ SNP array (Bianco et al. 2014)data y ¼ μ þ S þ ind þ fam þ cðÞfam þ e ijkl i j k kl ijkl from the FruitBreedomics project (Laurens et al. 2010, 2018) (data not shown). The reported causal SNP for the major pheno- where μ is the overall mean; S is a fixed year-replicate effect; i typic difference at the Ma locus, Ma1-SNP1455 (Bai et al. 2012) ind is the random additive effect of individual, with ind∼NID ÀÁj , was also used to genotype the germplasm set using Taqman® ; σ2 0 aA where A is the matrix numerator relationship based on assay AHCTENE (Chagné et al. 2019); this SNP was used to genome-wide inbreeding coefficients calculated from pedigree confirm results from the QTL analyses. Positions of markers on ð ; σ2 information and where NID 0 a ) means normally and inde- the GDDH13v1.1 reference genome were determined from the σ2 pendently distributed with mean 0 and variance ; fam is the Genome Database for Rosaceae (Jung et al. 2014)using ∼ ; σ2 BLASTn or, in cases of no hits, through manual examination family random effect, with fam NID 0 f ; c(fam) is the ran- ÀÁof this genome sequence through the genome browser at https:// dom effect of individuals within families, with c∼NID 0; σ2 ; ÀÁ o iris.angers.inra.fr/gddh13/ Bvisit the apple genome.^ ∼ ; σ2 and e is the residual term, with e NID 0 e . Narrow-sense 2 2 heritability (h ) and broad-sense heritability (H ) were calculated QTL detection and characterization from the following expressions: FlexQTL™ software, which implements pedigree-based QTL h2 ¼ σ2=σ2 a p analysis via Markov Chain Monte Carlo (MCMC) simulation 2 ¼ σ2 þ σ2 þ σ2 =σ2 H a f o p (Bink et al. 2002, 2014; www.flexqtl.nl), was used for QTL 2 ¼ σ2 =σ2: analyses, where an initial genome-wide analysis was followed D 4 f p I 2 ¼ σ2−3σ2 =σ2 by an analysis targeting only the set of LGs that harbored o f p significant and repeatable QTLs. The model parameter set- σ2 σ2 σ2 tings (Online Resource 7) enabled convergence to be reached. where a is the additive variance, f is the family variance, o is The Bayes factor parameter (2lnBF10) was interpreted as non- the variance component associated with individuals within fam- significant (0–2), positive (2–5), strong (5–10), or decisive (> σ2 2 ily replicated over years, e is the total residual variance, and D 10) evidence for presence of QTLs (Kass and Raftery 1995). and I2are dominance and epistasis effects, respectively, Chromosome-wide evidence was collected for NQTL models expressed as proportions of total phenotypic variance (Costa e 1/0 and 2/1 to nn− 1. QTLs were considered significant and Silva et al. 2004). Total phenotypic variance was calculated from repeatable where the 2lnBF10 for QTLs in the same genomic the observed phenotypic data and set equal to the above men- region for at least most of the nine evaluation points were σ2 ¼ σ2 þ σ2 þ σ2 þ σ2 tioned variance components as p a f o e ,and equal to or greater than 5. For data of each storage period, σ2 ¼ σ2−σ2−σ2 −σ2 the residual variance was calculated as e p a f o. replicated runs were performed until two runs were obtained Tree Genetics & Genomes (2019) 15:18 Page 5 of 17 18 that met the effective chain size criteria for convergence. QTL (p < 0.05). Families derived from selection W1 were excluded regions were recorded as a series of successive 2 cM bins with from such quantitative analyses of Ma alleles because of func- 2*ln Bayes factors that were greater than 5. MapChart tional diversity detected between its Ma3 alleles. To examine (Voorrips 2002) and VisualFlexQTL™ (www.flexqtl.nl) differences in effect among QTL alleles at Ma3 for W1, W1 software were used to visualize FlexQTL™ output on traces offspring were categorized according to their SNP haplotype of convergence of QTL models, positions, and genotypes. The at Ma3. Each of the two sets was further subdivided into three mode of a QTL was used as its most probable position. classes according to their FlexQTL™-estimated number of Delineation of QTL regions, QTL genotypes (QQ, Qq,or Bbackground^ q alleles at Ma and Ma3. Next, TA values were qq), breeding values, and additive variances associated with normalized by dividing an individual’sobservedTAbythe the QTL regions were estimated with FlexQTL™,where average TA of the offspring with the same number of back- alleles Q and q refer to alleles associated with high and low ground Q alleles. Effects of the two W1 haplotypes were then phenotypic values, respectively. compared across each background using the two-sample The QTL analyses of FlexQTL™ aggregates segregation Kolmogorov-Smirnov test (p < 0.05). Representation of mark- information across SNPs. It translate this information to IBD er haplotypes in the breeding families was calculated based on probabilities of 2 cM bins (chromosome segments), whereby FlexQTL™-derived identity-by-descent (IBD) estimates (re- founder alleles are used as statistical factors. This aggregation ported in FlexQTL’soutputfileBpedimap.ped^). of segregation information makes the analyses less sensitive Dominance levels of each QTL were calculated from the for individual markers having (some) missing data. To facili- difference between TA content of Qq heterozygotes and the tate downstream analyses by other software, we imputed miss- mean of the qq and QQ homozygotes, expressed as its pro- ing data for the QTL-associated SNP markers where possible. portion of half the difference in TA content between the two Original and missing-data-curated genotypic data are provid- homozygotes. The level of dominance of each QTL was esti- ed in Online Resource 6. mated for each storage period and year on subsets of offspring The proportion of phenotypic variance explained by each that had the same QTL genotype for the other, non-targeted QTL was estimated from FlexQTL™ output of an additive QTL and that also had at least three individuals representing genetic model by dividing the variance explained by the each of the three QTL genotype classes. In case of inconsis- QTL region (AVt1) by the total phenotypic variance for each tencies, the most prevalent genotype was used. These obtained run replicate (similar to Mangandi et al. (2017) but in that consensus QTL genotypes were used in each of the nine stor- paper the QTL variances were incorrectly weighted by the age period-year combinations. probability, the proportion of sampled QTL models that upheld the QTL; as the QTL reported there had a probability close to 1, the numerical difference is minimal). These estimates were Results conducted only where the QTL genotypes of full-sib family parents were identical to consensus QTL genotypes across all Acidity changes over storage and stability over years years and run replicates determined for the at-harvest data. The SNPs chosen for haplotyping each QTL were those In the original phenotypic data set, differences between years within a region representing a consensus of the highest poste- in TA at harvest were as high as 4.2 mg/L within the same rior densities determined by FlexQTL™ for each storage individuals. After conversion of at-harvest data to missing for period-year-run replicate QTL analysis. At the Ma3 QTL, 22 offspring deviating by at least 2.0 mg/L for 1 year from the three SNPs (ss475876134, ss475882878, and ss475879095) other two, higher correlations between years were achieved were removed that had a redundant segregation pattern across with r2 increasing from 0.54–0.66 to 0.65–0.78 (Online all germplasm to other SNPs (ss475882875, ss475878886, Resource 8). After this phenotypic data curation, across all and ss475879097, respectively). Haplotypes were constructed progenies and years, TA decreased by an average of 2.5 mg/ across the germplasm for Ma and Ma3 using PediHaplotyper L over the entire storage and ripening period (Fig. 1). For most (Voorrips et al. 2016), including automated imputation of offspring, about half this acidity loss had occurred after the missing data, with marker phasing by FlexQTL™. mid-duration storage treatment of 10wk (Fig. 1), such that for To examine differences in effect among Q alleles of differ- each year the three data points for the three storage periods ent origin, offspring were assigned to compound Ma and Ma3 were tight to their regression line (Online Resource 2). Within QTL genotypes such that comparisons could be based on in- each family, TA distributions tended to be continuous at har- dividuals with the same genotype for the second acidity QTL. vest, and the prevalence of discontinuous distributions in- Cumulative frequency distributions for TA content at harvest, creased with storage period (Online Resource 9). using averages for each of the 3 years, were generated and Broad- and narrow-sense heritability estimates decreased year-pairs of distributions were statistically tested for differ- with storage duration, from 0.87 and 0.84 respectively at har- ences with the two-sample Kolmogorov-Smirnov test vest to 0.63 and 0.22 for 20wk (Table 1). The calculated 18 Page 6 of 17 Tree Genetics & Genomes (2019) 15:18

Fig. 1 Boxplot distributions of titratable acidity measured for each of 3 years (2010–2012) at three storage durations (H = at harvest with no storage; 10wk = after 10 weeks of cold storage followed by 1 week at room temperature; 20wk = after 20 weeks of cold storage followed by 1 week at room temperature) on 243 apple germplasm individuals representing nine important breeding parents (listed in Table 2)

variance components of the broad-sense heritability associat- by approximately 5 cM among years (Fig. 2) and were consid- ed with family and individuals within families were minor for ered to represent the same QTL. The position mode of the LG 8 H and 10wk, thus supporting use of an additive genetic model QTL was at 28 cM for two of the years (2010 and 2012), and the in QTL analyses. However, for 20wk, these two variance frequency by which this position occurred in the sampled QTL components and additive variance had similar contributions models dominated that of any other position. In 2011, the QTL and each had large standard errors. interval was more dispersed but included the consensus posi- tion. For the LG 16 QTL in 2010 and 2011, FlexQTL™ results suggested it was positioned in the first 4 cM of the chromosome, QTLs detected: Ma and Ma3 whereas for 2012 it tended to be detected in the 6–8cMbin. These LG 16 QTL positions coincided with the genetic position For each storage treatment and year combination, two to six of the previously reported Ma (Maliepaard et al. 1998; Liebhard simulations with different seeds were needed to obtain full con- et al. 2003;Khanetal.2013). Furthermore, two SNPs in the 6– vergence in two of those simulations that were then designated 8 cM bin, ss475881682 and ss475876558, flanked the as two replicate runs. At harvest, two QTLs for fruit acidity with aluminum-activated malate transporter gene ALMT1 decisive statistical evidence (2lnBF10 > 30) were detected at LG (MDP0000252114, Online Resource 11), previously deter- 8 and LG 16 in each year and with each replicative run, and no mined to likely represent the gene underlying Ma (Bai et al. further significant QTLs were found (Online Resource 10). 2012;Khanetal.2013;Maetal.2015b). Hereafter, the LG Genetic positions of the mode of each of these two QTLs varied 16 QTL is considered to be Ma. The LG 8 QTL is hereafter referred to as Ma3 (BMa2^ was previously assigned by Bai et al. Table 1 Heritability estimates for titratable acidity (± standard errors) at (2012) to a candidate gene in the vicinity of Ma). three storage periods across 3 years. H2 and h2 are broad- and narrow- sense heritability estimates, respectively, and D2 and I2 are family and QTL analyses for fruit acidity after storage detected QTLs at individuals-within-family components of H2, respectively the same genomic regions as for at harvest, i.e., Ma and Ma3 (Fig. 2), and no additional significant QTLs were found (Online H2 h2 D2 I2 Resource 12). Evidence for the presence of these after-storage H 0.87 ± 0.02 0.84 ± 0.08 0.05 ± 0.10 − 0.01 ± 0.14 QTLs was more variable than for acidity at harvest although the 10wk 0.73 ± 0.05 0.61 ± 0.18 0.07 ± 0.13 0.04 ± 0.19 posterior probability remained strong to decisive, except in one 20wk 0.63 ± 0.05 0.22 ± 0.20 0.22 ± 0.23 0.19 ± 0.21 case (20wk in 2011) where the evidence was just below the threshold for both run replicates (Online Resource 12). Tree Genetics & Genomes (2019) 15:18 Page 7 of 17 18

Table 2 Summarized interpretation of Ma and Ma3 genotypes for important breeding parents. Full details for each of the two run replicates for each of three storage treatments and each of 3 years of phenotypic data are provided in Online Resource 13. Alternative Q alleles (Q2) are described in the text

Important breeding parent QTL genotype

Ma Ma3

Arlet Qq qq Aurora Golden Gala Qq qq Cripps Pink Qq Qq Delicious Qq qq Enterprise Qq Qq

Honeycrisp QQ2 qq Splendour Qq qq WA 5 Qq Qq

W1 QQ Q2q

(Online Resource 13). Ma and Ma3 genotype estimates for ‘Enterprise’ and W1 were sometimes inconclusive at 10wk or 20wk. Consistent after-storage QTL genotype estimates for those two parents could not be identified from alternative parameter settings nor additional data curation considerations. The final interpretation of QTL genotypes for each of the nine parents (Table 2) was that determined from the at-harvest re- sults. Although zero to four Q alleles were possible within any individual across the two loci, all parent cultivars and selec- tions had only one (‘Delicious,’‘Arlet,’‘Aurora Golden Gala,’ and ‘Splendour’) or two (all others) Q alleles. Fig. 2 Posterior intensities of acidity QTLs at Ma and Ma3 QTLs along their corresponding linkage groups (LGs) for three storage treatments. Sources, associated SNPs, and effects of QTL alleles Panel a = H treatment, at harvest. Panel b = 10wk treatment (10 weeks c of cold storage followed by 1 week at room temperature). Panel =20wk Ancestral origins of the parental Q and q alleles were success- treatment (20 weeks of cold storage followed by 1 week at room temper- ature). The x-axis represents combined genetic locations of LG 16 (blue fully traced via SNP haplotypes spanning the QTLs (Table 3). lines, Ma) and LG 8 (pink lines, Ma3) in centimorgans (cM). The y-axis For Ma, each parent of the full-sib families had one or two Q represents posterior intensities based on an additive genetic model exe- alleles that were traced through the pedigree to eight or nine ™ cuted by FlexQTL software. Increasing color grades correspond to ancestors: ‘’‘Frostbite,’‘Duchess of Oldenburg,’ NJ successive years. Solid and dotted lines represent the two run replicates within each year’sruns 27, NJ 136055, perhaps F2-26829-2-2, and the unknown pa- ternal parents of ‘,’‘Jonathan,’ and ‘.’ Ancestors ‘,’‘McIntosh,’ and an QTL genotypes of important breeding parents unknown parent of ‘Delicious’ were the sources of q alleles for the Ma QTL. For Ma3,theQ alleles for high acidity in Genotype estimates revealed that the Q allele representing important breeding parents ‘Cripps Pink,’‘Enterprise,’ and relatively high-acidity levels was common among breeding ‘WA 5’ were inherited from ancestors ‘,’ parents for Ma and rare for Ma3 (Table 2). For Ma, QTL ‘McIntosh,’ and F2-26829-2-2, respectively. genotypes for TA at harvest were mostly consistent across Specific SNP markers were indicative of the high-acidity years and run replicates (Online Resource 13); two parents Ma allele. The B-allele of SNP marker ss475882553 was al- were classified as QQ (‘Honeycrisp’ and W1), while the other most always associated with Q-Ma among the parental haplo- seven were classified as Qq. For Ma3, QTL genotype esti- types. Only one Q-allele source (‘Duchess of Oldenburg’) mates for TA at harvest were also mostly consistent across lacked this SNP allele, but a unique B-allele at another SNP years and run replicates; three were classified as Qq (‘Cripps (ss475881815) was source-specific for this Q allele. The pres- Pink,’‘Enterprise,’ and ‘WA 5’) and the other six were qq ence of two distinct Q-allele sources for Ma was thus 18 Page 8 of 17 Tree Genetics & Genomes (2019) 15:18

Table 3 QTL genotypes of Ma and Ma3 for nine important breeding those in parentheses were not included in the pedigree input data for QTL parents, with associated SNP haplotypes and ancestral origins. QTL discovery. Evidences for the alternative Q2 alleles are described in the alleles are presented in order of maternal (top) and paternal (bottom) text. Haplotypes with the same SNP allele sequence are given the same origin for each parent cultivar. SNP order for sequence of alleles follows letter label; where such haplotypes traced to different founders, subscripts the genetic map order provided in Online Resource 11.TheB-alleles of indicate the founder source, and those with different Q/q effect designa- the second and fifth SNPs of Ma, ss475881815 and ss475882553, respec- tions are indicated by their suffix. At Ma3,theB haplotype of Honeycrisp tively, always associated with Q alleles, are shown in bold. Ultimate was determined to be derived from a recombination between founder ancestral sources are shown in bold italics at the end of the list of succes- haplotypes (E-ma3 of Grimes Golden and a unique haplotype of un- sive ancestors through which the IBD alleles could be traced, although known effect from Duchess of Oldenburg)

QTL Important breeding QTL allele SNP haplotype Successive ancestors parent Allele sequence Label

Ma Honeycrisp Q2 ABBAABBBAAA A-Ma MN 1627 > Duchess of Oldenburg

Arlet Q BAAABBAAABA BF2/Jt-Ma > (Jonathan, F_Jonathan) Enterprise Q BAAABBAAABA Co-op 7 > (PRI 1018-9 > PRI 47-147 > F2-26829-2-2 or Jonathan > F_Jonathan)

Aurora Golden Gala Q BAAABBAAABA BWs-Ma Splendour > Delicious > (Winesap) Delicious Q BAAABBAAABA (Winesap) Splendour Q BAAABBAAABA Delicious > (Winesap) Cripps Pink Q BABBBABAABA C-Ma Lady Williams > (F_Lady Williams) Honeycrisp Q AABBBBBBBBB D-Ma Keepsake > Frostbite WA 5 Q BAAABABAABA E-Ma Co-op 15 > NJ 27 W1 Q AABABBAAABA F-Ma NJ 90 > (NJ 136055) W1 Q AABABBBAABA G-Ma Goldrush > Golden Delicious > (Grimes Golden) Arlet q AABAABBBBAB H-ma Golden Delicious > (F_Golden Delicious) Cripps Pink q AABAABBBBAB Golden Delicious > (F_Golden Delicious) Splendour q AABAABBBBAB Golden Delicious > (F_Golden Delicious) WA 5 q AABAABBBBAB Splendour > Golden Delicious > (Grimes Golden) Aurora Golden Gala q BAAAABBBBAB I-ma Gala > Kidd’s Orange Red > Delicious > (F_Delicious) Delicious q BAAAABBBBAB (F_Delicious) Enterprise q BAAAABBBAAA J-ma PRI1661-2 > McIntosh Ma3 Cripps Pink Q BAAABAAAA A-Ma3 Lady Williams > Granny Smith

Enterprise Q BABBABBBA BMc-Ma3 PRI1661-2 > McIntosh hap 1

WA 5 Q BAAABAAAB CF2-Ma3 Co-op 15 > PRI 612-1 > (PRI14-126 > F2-26829-2-2)

W1 Q2 ABBABBABB D-Ma3 NJ 90 > > McIntosh hap 2 Arlet q BABBABBBB E-ma3 Golden Delicious > (Grimes Golden)

Honeycrisp q BABBABBBA BGG+ MN 1627 > Golden Delicious/Duchess of Oldenburg > DO-ma3 (Grimes Golden/Duchess of Oldenburg) Aurora Golden Gala q BAAABBABB F-ma3 Splendour > Golden Delicious > (F_Golden Delicious) Cripps Pink q BAAABBABB Golden Delicious > (F_Golden Delicious) Splendour q BAAABBABB Golden Delicious > (F_Golden Delicious) WA 5 q BAAABBABB Splendour > Golden Delicious > (F_Golden Delicious) W1 q BAAABBABB Goldrush > Golden Delicious > (F_Golden Delicious)

Arlet q BAAABAAAB CES-ma3 Idared > Jonathan > () Aurora Golden Gala q ABBABAABA G-ma3 Gala > Kidd’s Orange Red > Delicious > (F_Delicious) Delicious q ABBABAABA (F_Delicious) Splendour q ABBABAABA Delicious > (F_Delicious) Delicious q BAAABAABB H-ma3 (Winesap) Enterprise q ABBABAAAB I-ma3 Co-op 7 > (PRI 1018-9 > PRI 47-147 > Jonathan > F_Jonathan) Honeycrisp q AABBBBBBB J-ma3 Keepsake > indicated, a common one characterized by ss475882553-B those two SNPs were therefore always associated with a Q

(Q) and a rare one by ss475881815-B (Q2). The B-alleles for allele of Ma. For the reported causal SNP of the Ma locus, Tree Genetics & Genomes (2019) 15:18 Page 9 of 17 18

Ma1-SNP1455 via the Taqman® assay AHCTENE, the G (QQ/qq or QQ/Q2q) as a parent: ‘Cripps Pink’ (Qq/Qq)× allele was always associated with Ma,bothQ and Q2,and W1 and W1 × ‘WA 5’ (Qq/Qq). W1 segregated for its Ma3 correspondingly the A allele was always associated with q-ma. q alleles, D-Ma3 and F-ma3, which were respectively Some cultivars shared the same SNP haplotype for Ma3 while inherited without recombination by 13 and 29 offspring. these haplotypes came from different lineages and were associ- Each of the two sets was further subdivided into three classes ated with different-effect QTL alleles (Table 3). Haplotypes BMc- according to their number of Ma-Q and Ma3-Q alleles. The set Ma3 of ‘Enterprise’ and BGG+DO-ma3 of ‘Honeycrisp’ were of offspring having the D-Ma3 allele had a higher average TA identical-by-state (IBS) but not IBD according to available ped- content at H and 10wk with each number of Bbackground^ Q igree information, as ‘Enterprise’ inherited its haplotype from alleles and in each year (Online Resource 15). The distribu- ‘McIntosh’ while that of ‘Honeycrisp’ appeared to have arisen tions of the normalized TA values were statistically signifi- from a recombination event in its paternal parent MN 1627 (data cantly different at H and 10wk for each year (0.003 < not shown). The other case was haplotype CF2-Ma3 of ‘WA 5’ p < 0.047), except for H in 2011 (p = 0.38). The effect aver- (source: F2-26829-2-2) and haplotype CES-ma3 of ‘Arlet’ aged + 0.8 mg/L (Online Resource 15), which was less than (source: ‘Esopus Spitzenburg’). In this case, recent recombina- the effects of + 1.8 mg/L (H) and + 1.2 mg/L (10wk) for the tion within the haploblock appeared unlikely because haplotype- common Ma3-allele. Hence, W1 appeared to segregate for a sharing extended for 36 cM around this region (Online Resource Q2 allele with an intermediate effect between q and Q. 14). Lack of discrimination by the SNPs in the region also ap- peared unlikely because the phenomenon remained when using a Combined QTL effects higher resolution of 58 SNP markers from the 20K Infinium array (Bianco et al. 2014) within the same interval (data not Phenotypic variance explained by Ma, Ma3, and Ma+Ma3 shown). A comparison of harmonized TA data from subsets of was 26.9 ± 2.4%, 39.1 ± 0.6%, and 66.0 ± 1.9%, respectively, offspring that had otherwise identical QTL genotypes (while across the six storage-year-replication combinations for which distinguishing between Q and Q2 at both loci, results not shown) the estimated parental QTL genotypes were consistent with confirmed the statistical significance of the Q/q contrast for the the consensus genotypes (Online Resources 13 and 16). The

CF2-Ma3 and CGG+DO-ma3 haplotypes (p = 0.048). explained variance decreased with increasing storage duration The average effect of each Q-dose for each QTL across the for Ma (from 30.2 ± 2.1 for H to 20.5 ± 0.3% for 10wk) and 3 years was + 1.8 mg/L (H), + 1.2 mg/L (10wk), and + 0.9 mg/ remained stable for Ma3 (38.3 ± 0.3% for H and 40.7 ± 0.5% L (20wk). Loss of acidity over storage was greater for geno- for 10wk). Estimated dominance levels of Ma and Ma3 for TA types associated with higher acidity at harvest (i.e., acidity loss content across all years, run replications, and QTL genotypes 4× Q >3×Q >2×Q >1×Q >0Q). Extrapolated linear were 5.8 ± 1.1% and 1.5 ±1.9% respectively. regression lines indicated that, with a hypothetical storage Considering both QTLs at once, fruit acidity increased with treatment of B40wk^ (40 weeks cold storage and 1 week at Q-allele dose at each storage period (Fig. 3). A single Q-dose at room temperature), fruit acidity levels converged to a common Ma was associated with a higher TA content than was a single baseline level that depended on the year (Fig. 3). Loss of Q-dose at Ma3, adding 2.0 and 0.7 mg/L, respectively, at harvest acidity over storage was greatest in 2011 for all genotypes, (across all 3 years) compared to qq genotypes. In the presence of and this year had the lowest extrapolated baseline (Fig. 3c). atotaloftwoQ alleles across both loci, offspring heterozygous Averaged over the three observed years, the extrapolated base- at both loci had slightly higher or equal (but never lower) TA line was approximately 1.0 mg/L (Fig. 3d). content than offspring having both Q alleles at Ma (Fig. 3). To determine whether the two Ma-Q-allele types, common Comparisons to offspring with both Q alleles at Ma3 could not

Q and ‘Duchess of Oldenburg’–derived BQ2,^ were associated be made as such germplasm was not present in the dataset. In the with consistently significant differences in effect size, com- presence of a total of three Q alleles, offspring homozygous at parisons were made among offspring that had these different Ma3 tended to have a higher TA content than those homozygous Q-allele sources but the same genetic background at Ma3. at Ma. The effect of both loci being QQ-homozygous could not

These comparisons were among Ma-Qq, Ma-Q2q,andMa- be estimated due to too few such offspring. qq,withMa3 always being homozygous qq. A statistically significant difference in TA content was observed

(p < 0.005), whereby the rare ‘Duchess of Oldenburg’ Q2 al- Discussion lele was associated with a higher TA content, + 0.5 mg/L at harvest across 3 years, compared to the common, multi- Two large-effect QTLs influencing apple fruit acidity levels be- ancestor Q allele (Fig. 4). fore and after commercially relevant fruit storage treatment were Evidence was obtained for the presence in the selection W1 detected in US breeding germplasm. These QTLs were charac- of a Ma3 allele of higher acidity (Q2)withaneffectinbetween terized across breeding families derived from nine parents with that of the common q and Q alleles. Two families had W1 common ancestors but also distinct founders. The two QTLs 18 Page 10 of 17 Tree Genetics & Genomes (2019) 15:18

Fig. 3 Average fruit acidity content of Ma-Ma3 compound genotypes presence of one and two doses of the Q allele for Ma and Ma3 respec- over three storage periods observed for all three fruiting seasons (years) tively. QTL genotypes of offspring estimated with FlexQTL™ were together (a) and each year individually (b–d). Regression lines are ex- assigned from the at-harvest results and not considering the subsequently trapolated beyond the experimentally observed storage periods. determined Q2 alleles (thus, ‘Honeycrisp’ = QQ for Ma and W1 = qq for Compound genotypes are specified by two digits indicating the number Ma3 here) of Q alleles for Ma and Ma3 respectively. For example, B12^ indicates the that were consistently detected, Ma on LG 16 and Ma3 on LG 8 standardized manner from genetically variable populations (the latter locus being newly named here), represented loci de- (Evans et al. 2012;Howardetal.2018), the high heritability tected previously in biparental family studies. Combined effects estimates obtained indicate that any discrepancies in determin- of alleles from the two loci highlighted the utility of this DNA- ing standardized maturity across trees within and among years based information in new cultivar development for targeting were minimal for fruit acidity compared to genetic influences. desired fruit acidity levels before or after storage. Kouassi et al. (2009) reported narrow-sense heritabilities for four storage treatments of 0.79 ± 0.01 to 0.81 ± 0.01 for TA on General genetic influences on fruit acidity 2207 pedigreed individuals in 29 families from breeding pro- from harvest through storage grams of six European countries. Those heritability estimates were therefore stable over storage and were also the highest The high heritability estimates for apple fruit acidity at harvest among numerous fruit quality traits evaluated in that study, in (H2 = 0.87 ± 0.02) and for the 10wk treatment (H2 =0.73± contrast with our study’s decrease in this general genetic sig- 0.05), largely due to additive variance, across the 16 full-sib nal with storage. The difference between our results and those families from nine parents and 3 years of evaluation was as of the previously mentioned report might be due to differences expected from previous studies (Baojiang et al. 1995; in phenotyping or statistical methods or storage treatments. Maliepaard et al. 1998;Liebhardetal.2003; Kouassi et al. More likely, there is a fundamental difference in QTL geno- 2009). These high estimates signified a large opportunity to types and alleles present—for example, several ancestors of identify the underlying genetic factors segregating in this Kouassi et al. (2009) were associated with extreme breeding germplasm. While it is difficult to harvest apple fruit in a values for acidity—perhaps homozygous QQ for Ma3 and

Fig. 4 Cumulative frequency distribution of TA content at harvest for offspring differing for the source of their single Q allele at Ma while having the same genotype (qq)atMa3 Tree Genetics & Genomes (2019) 15:18 Page 11 of 17 18

Ma, or having BQ3^ alleles associated with extremely high being explained by a LG 16 QTL (Ma), similar to above, but acidity—and germplasm of these types were not represented only 8% by a LG 8 QTL (Ma3). That lower influence of Ma3 in the present study. The low additive genetic variance calcu- canbenowexplainedbythesegregationofaQ2 allele associ- lated for the 20wk treatment (Table 1) indicated limitations for ated with acidity levels not as high as the common Q allele. an additive genetic model to adequately describe the pheno- Ma is reported to have additive gene action in papers where typic data. Reanalyzing the 20wk dataset with a genetic model titratable acidity was used as the measure of fruit acidity in FlexQTL™ that allowed for the presence of both additive (Liebhard et al. 2003), and dominant gene action was reported and dominance effects did not find dominance to be a signif- where pH was used (Maliepaard et al. 1998;Xuetal.2012; icant effect for Ma and Ma3 (results not shown). The differ- Khan et al. 2013). These conclusions appear to be conflicting, ence between the univariate analysis results, which indicated but might be explained by differences in the scales of observa- significant dominance effects for 20wk, and the FlexQTL™ tion, TA being linear and pH being logarithmic. In QTL evidence showing otherwise might be due to dominance ef- discovery approaches that assume additive gene action, such as fects contributed by many small-effect trait loci other than Ma in our current analysis, the use of TA would therefore be the and Ma3 and thus were not detected by FlexQTL™. appropriate measure of acidity. The appropriate QTL analytical model for sensory evaluations of acidity would depend on Identities and effects of detected QTLs whether the scale was more aligned with TA or pH. In the study of Jia et al. (2018) where malic acid content was assessed The two detected QTLs influencing apple fruit acidity for all by HPLC, genetic action of the Ma allele varied between full-sib storage treatments were the only ones that had statistical evi- families, being dominant in ‘Jonathan’ × ‘Golden Delicious’ and dence. The detection of these two QTLs was consistent with additive in Malus asiatica ‘Zisai Pearl’ × M. domestica ‘Red previous reports on biparental families using fruit evaluated .’ The cause for this variable genetic action is not clear. only at harvest. Our detected BMa^ locus was very likely that controlled by the aluminum-activated malate transporter Two-allele model and beyond ALMT1 gene on LG 16 described by Bai et al. (2012)and Khan et al. (2013), as evidenced by genomic positioning. In QTL modeling, FlexQTL™ assumes the presence of just two Our BMa3^ might be identical to the acidity QTL reported functional QTL alleles (Q and q), associated with high and low on LG 8 by Liebhard et al. (2003), Kenis et al. (2008), phenotypic values respectively. These associations are based on Zhang et al. (2012), and Ma et al. (2015b) based on co- phenotypic contrasts among Q and q alleles determined to be localization on aligned genetic linkage maps and/or the different from IBD analysis across the pedigree structure. GDDH13 reference genome sequence (Online Resource 17). Observations of extended shared haplotypes across and around Effects of the two QTLs detected were maintained across the QTL interval and other chromosomal regions (results not many genomic backgrounds typical of breeding germplasm, shown) indicate a recent shared, still unknown, ancestor, such providing confidence in breeding consideration of genotypes as for BF2/Jt-Ma from F2-26829-2-2 or the unknown father of for these QTLs. Our 9-parent, 16-family germplasm set had a ‘Jonathan’ and BWs-Ma from ‘Winesap’ (Table 3). However, as wider range of genetic backgrounds than the biparental fami- QTL alleles calculated to be functionally equivalent arose from lies of Liebhard et al. (2003), Zhang et al. (2012), and Ma et al. numerous ancestral sources (founders) (Table 3), some of these (2016)(‘Fiesta’ × ‘Discovery,’‘Jonathan’ × ‘Golden sources might actually have a different QTL allele. Indeed, two

Delicious,’ and ‘Jiguan’ × ‘Wangshanhong,’ respectively), extended shared haplotypes at Ma3 andbeyond(CF2-Ma3 and allowing for more allelic combinations of the two large- CES-ma3, Online Resource 14)hadcontrastingQ/q designa- effect QTLs and other influencing loci. Our large and similar tions. As the contrasting Q/q designations were confirmed by effects calculated for the Ma and Ma3 QTLs were in agree- comparing offspring with otherwise identical QTL genotypes ment with those previous biparental studies. The phenotypic (p = 0.01), a mutation from Q to q or vice versa is likely to have variance explained for each (Ma, 27 ± 2% and Ma3, 39 ± 1%, occurred and might be of relatively recent origin. In cases such with a combined effect of 66 ± 2%) was slightly less than as both C F2-Ma3 and CES-ma3 segregating, diagnostic DNA Liebhard et al. (2003) reported (42% and 46%, respectively), tests will need to ascertain lineage-specific effects or IBD esti- higher than reported by Ma et al. (2016) (12% and 14%, re- mates prior to use in marker-assisted selection. spectively) and Zhang et al. (2012) (13.5% for Ma3) for their ItappearslikelythateachofthetwoQTLshavemorethan single Ma3. It is noteworthy that QTL effects were calculated two-allele types, as represented by the Q2 alleles. The higher- to be so high in the present study, indicating that the two QTLs than-high-acidity Ma-Q2 allele of ‘Honeycrisp,’ inherited from indeed account for the bulk of genetic variation for fruit acid- ‘Duchess of Oldenburg,’ had a distinctive SNP haplotype in- ity across a wide range of breeding germplasm. Kenis et al. cluding a unique allele for the second SNP. The reported causal (2008), using another biparental family (‘Telamon’ × SNP for the Ma locus, Ma1-SNP1455 (Bai et al. 2012;Chagné

‘Braeburn’), reported 30–34% of the phenotypic variation et al. 2019) was not able to distinguish the Ma-Q2 from Ma-Q. 18 Page 12 of 17 Tree Genetics & Genomes (2019) 15:18

However, a combination of any two of the three SNPs of Ma1- segregating from ‘Jonathan’ (Zhang et al. 2012; Sun et al. SNP1455, ss475881815, and ss475882553 would be effective 2015; Jia et al. 2018). Assuming the ‘Jonathan’ used in these to do so (and to establish a germplasm individual’s Q-allele studies was true-to-type, and if titratable acidity and HPLC- dosage). ‘Duchess of Oldenburg’ is an old Russian cultivar with assessed malate content are equivalent for apple fruit, then the origins distinct from the bulk of European and US apple germ- effect of Ma3 might be sensitive to environmental factors and/ plasm (Beach et al. 1905), so the unique effect of its allele is not or genetic background. surprising. Assuming that the high-acidity Q allele is ancestral to low-acid q allele contributing to increased animal palatability Diagnostic markers and causal genes (Ma et al. 2015a;Duanetal.2017), the seniority, genesis, and

DNA sequence differences of Ma-Q2 and the common Ma-Q Two or more of the available SNP markers for each locus are yet to be clarified. would be required to diagnose the alleles present in any indi-

Ma3-Q2 had low representation as it was present in just two vidual. For the Ma locus, the SNP marker ss475882553 full-sib families that also showed distorted segregation. (GDsnp01588) was predictive for Q/q from ten of the eleven Therefore, its discovery was at risk of being false, being due ancestral sources. Its predictiveness is further confirmed by to the coincidental more than usual presence of phenotypic consistency of its QTL genotypes in parents of reported outliers. Confirmation on it being true comes from its co- QTL studies on acidity. For example, ‘Braeburn,’ localization with a QTL of relatively small effect reported by ‘Discovery,’‘Fiesta,’‘Prima,’ and ‘Telamon’ were reported Kenis et al. (2008). That reported result can now be explained to be heterozygous at Ma (Maliepaard et al. 1998;Liebhard by both parents in that study being Q2q rather than qq,thus et al. 2003; Kenis et al. 2008) and were all heterozygous for having a smaller phenotypic contrast between the QTL’sseg- this marker (in the present dataset or that of FruitBreedomics). regating alleles than would Qq parents. Like W1 in the present Only for ‘Duchess of Oldenburg’ and derived germplasm car- study that inherited the Ma3-Q2 allele from ‘McIntosh,’ the rying the Q2 allele from this source was ss475882553 not mapping family’sparent‘Telamon’ in Kenis et al. (2008)is predictive. However, the nearby SNP marker ss475881815 also a (direct) ‘McIntosh’ descendant and the allele it inherited alone discriminated this rare Q-allele source. Statements on from ‘McIntosh’ was determined by us, using its predictiveness in segregating families require insights on its

FruitBreedomics SNP data (results not shown), to be IBD with distance from the actual gene to which this Q2 allele belongs. the W1 allele. Similarly, the second allele of ‘Telamon’ If this would indeed be the ALMT1 gene of Ma, the physical inherited from ‘Crimson Golden’ through ‘Golden distance is more than 3 Mb, suggesting a genetic distance of Delicious’ was part of haplotype E-ma3 (Table 3). The exis- approximately 7 cM. Also, further information on the strength tence of multiple degrees of Q-allele effects that were well of the currently found association would be required prior to represented in the present study’s germplasm could explain its application on wider germplasm. why FlexQTL™ was not able to reach convergence in its The reported causal polymorphism for Ma/ma is based on QTL genotype estimates for W1 in some runs for 2010. an intragenic SNP creating a truncated, ineffective protein for Use of the IBD approach (Bink et al. 2014) might be useful the ma allele, providing the basis for a functional diagnostic to infer QTL genotypes of other breeding parents, ancestors, DNA marker (Ma et al. 2015b; Jia et al. 2018). However, it and founders that are genetically related to our currently ex- does not appear that ‘Duchess of Oldenburg’ or its descen- amined important breeding parents. Such inferences would dants were included in the study that identified that causal help extend QTL genotyping results on specific germplasm mutation, allowing for the possibility that this cultivar carries to wider germplasm. Comparisons among QTL studies could a different mutation at the Ma locus or in a nearby gene such provide insights into the robustness of QTL effects across as the Ma2 gene described by Bai et al. (2012). Because none genetic backgrounds and environments. For example, the of the current single SNP markers at the Ma3 locus were presence of a Ma3-Q2 allele in our study was consistent with predictive for any Q/q source across the germplasm (they were another QTL mapping study (Kenis et al. 2008)ongermplasm only predictive within families), extrapolating the Q/q status that had a common founder origin of the favorable QTL allele. of Ma locus alleles across germplasm requires consideration In addition to confirmations, we also noticed some of multiple SNPs at once, i.e., haplotypes. inconsistencies. Sun identified 194 genes on the GD 1.0 draft genome se- Our data allowed complete QTL genotyping of ‘Jonathan’ quence corresponding to the LG 8 QTL found in ‘Jonathan’ × and ‘Golden Delicious,’ two major founders in apple breeding ‘Golden Delicious,’ and performed some initial candidate worldwide. These founders were predicted to be heterozygous gene expression studies. Results from qPCR, complementa- for Ma and homozygous qq for Ma3.Incontrast,Ma seemed tion, transient expression, and protein studies reported by Jia not to segregate in studies of a ‘Jonathan’ × ‘Golden et al. (2018), continuing the work of Sun et al. (2015), gave Delicious’ full-sib family with acidity measured as malate insight into the gene networks and signaling pathways in- content via HPLC, whereas a QTL was found for Ma3 volved in vacuolar malic acid content: MdPP2CH inactivated Tree Genetics & Genomes (2019) 15:18 Page 13 of 17 18 three vacuolar H+-ATPases that serve as proton pumps across presence of outliers might raise uncertainty on the true pres- the tonoplast (MdVHA-A3, MdVHA-B2, and MdVHA-D2)as ence of multiple Q alleles, especially for Ma3 as its discovery the aluminum-activated malate transporter MdALMTII. Also, was based on a low number of offspring. In that respect, it was

MdPP2CH was determined to be suppressed by the early aux- useful that our findings on Ma3-Q2 were consistent with pre- in response gene, MdSAUR37. Jia et al. (2018) focused on vious research of Kenis et al. (2008). QTL LG 8 co-localizing genes that showed the highest differ- Another limitation was in the bi-allelic nature of the QTL ential segregation between two bulks of high- and low-malate models used in QTL mapping. In the presence of multiple Q offspring and prioritized genes that were segregating in both alleles of different effect, FlexQTL™ could be wrong in its Q the parents ‘Jonathan’ and ‘Golden Delicious,’ despite the LG vs. q designations. The presence in the germplasm of more 8 QTL segregating in ‘Jonathan’ only. These prioritized genes than one kind of high-acidity Q allele for Ma and Ma3,with showed a distorted ratio of zygotic genotypes among the 246 quantitatively differing effects for each, might have contribut- randomly chosen offspring of ‘Jonathan’ × ‘Golden ed to the software’s inability to determine consistent QTL Delicious.’ For example, their MdPP2CH marker showed a genotype estimates for certain parents and might have led to AA:AG:GG segregation of 82:97:67, a significant deviation incorrectly designated alleles for others. For example, the from the expected 1:2:1 ratio (p = 0.002), and showed a sig- presence of the Ma-Q2 allele in ‘Honeycrisp’ might have hin- nificant lack of heterozygotes (p = 0.005). The presence of dered FlexQTL™’s ability to accurately designate Q vs. q distorted segregation for the LG 8 QTL region in the randomly alleles for that cultivar for both QTLs. This possible issue chosen offspring indicates the presence of selection pressure had only a minor role in our study. A co-factor analyses with against certain alleles or zygotes. Depending on how this SNP ss475882553 as a co-factor representing Ma-Q2 did not selection-sensitive trait co-segregated with malate content, affect the previous consensus QTL genotypes—with one ex- the differential segregation between the bulks might have aris- ception (Online Resource 18). Previously, W1 was assigned en not only from artificial selection for contrasting malate Ma3-qq, whereas in the co-factor analyses results were incon- content. This feature might make the bulked segregant analy- clusive: estimates were inconsistent between years and even sis approach that was used ineffective for the intended pur- within a single simulation (2012-10wk, R2, Online Resource pose. An alternative, efficient way to identify candidate genes 18). Accounting for the higher-than-high-acidity Ma-Q2 allele and develop predictive markers for Ma3 might be selective increased awareness of QTL genotype assignment issues with phenotyping and intense genotyping of offspring with recom- W1, which we can now understand from the presence of a bination in the QTL interval. Segregation distortions in the less-strong Ma3-Q2 allele in this selection. randomly chosen offspring might indicate the presence of le- While the previous two limitations related to confidence thal alleles (Gao and van de Weg, 2006). If true, the observed in the discovered QTL models, a third limitation relates to zygotic genotype ratio was consistent with biparental segre- obtaining practical insight from the implications of our QTL gation of a single, recessive, lethal allele for a locus located model. Under-representation in some parts of the compound approximately 10 cM from the LG 8 QTL, for which the lethal QTL genotypes (Online Resources 19 and 20)hampers allele of ‘Jonathan’ was in coupling phase with the Q-allele for strong conclusions on the performance of QTLs across wide malate content (data not shown). germplasm. Under-representation is true for most QTL map- ping studies on a multi-locus trait. In our study, classes with Limitations in this study three to four Q-doses were mostly under-represented. This outcome is readily expected from binomial probability. This study observed and managed large experimental varia- Assuming regular segregation and a balanced representation tion in the phenotypic data. While TA content generally de- of QTL genotypes across the parent pairs, the class of creased during storage, some samples showed a large increase smallest sample size theoretically holds 1/16 (6.3%) and 1/ of up to 3.7 mg/L. Also, individuals could show large differ- 81 (1.2%) of the total population in the case of a bi-allelic or ences between years in TA at harvest. Such phenomena have tri-allelic two-locus trait, respectively. With a population size not been reported in previous QTL mapping studies, which of around 160 phenotyped offspring per dataset such as in might indicate previous ignorance or the presence here of a our study, such a class would hold 10 and 2 offspring, re- more than usual level of experimental error. Curation for ex- spectively. In addition, QTL genotypes are usually not treme outliers improved the between-year correlations, al- equally distributed over parent pairs and sampling errors though still only to moderate levels (Online Resources 3, 4, occur also, increasing chances for genotype classes with no 5,and8), probably leaving considerable experimental varia- or very low representation. However, because of obtained tion in the genetic analyses. This residual variation might ex- insights on effects of trait loci, numbers of alleles, and pa- plain the variability in the observed mode and interval size of rental QTL genotypes, directed crosses can now be made for the QTLs and in the parental QTL genotype estimates among future studies to examine the effects of particular compound storage treatments and years (Fig. 2, Online Resource 13). The QTL genotypes. 18 Page 14 of 17 Tree Genetics & Genomes (2019) 15:18

Q-allele dosage model of apple fruit acidity conditions over several years. It would also be valuable to identify pre-harvest factors that influence initial acidity levels The presence of just two major QTLs and the consideration and final baselines both for prediction purposes and to inform of just two alleles per locus provided a simplified interpre- management practices that could help achieve fruit within tation of fruit acidity genetic control that might lend itself target acidity levels for as long as possible to extend market to easy integration into routine breeding decisions. The life. largely additive genetic variation, compared to relatively low dominance and epistasis, and the similar-sized effects of the two QTLs, as observed for the H and 10wk evalu- Breeding targets for Q-allele dosage ations, indicated that the dosage of high-acidity alleles could be a simple way to understand effects of the two Our results indicate that breeders can select for particular Q- loci. In this model, as long as high-acidity alleles can be allele dosages to achieve a particular desired level of high distinguished and counted for any individual, there is no acidity after storage—such as doses of 3× and 4× to achieve need to consider the particular allelic combinations. Some fruit with 5 mg/L TA after a 20wk treatment (Fig. 3a). interesting features arose from this model, displayed in Similarly, the results indicate that, to avoid acidity being too Fig. 3. First, at harvest were observed the largest phenotyp- low, particular Q-allele dosages should be avoided—such as ic distinctions among Ma-Ma3 compound genotypes, which 0× or 1× doses to avoid fruit with less than 3.5 mg/LTA after a went along with the highest broad- and narrow-sense heri- 20wk treatment (Fig. 3a). As acidity levels of > 5 mg/L are tability estimates (Table 1). Second, the higher the Q-allele generally considered desirable for apple fruit (Harker et al. dosage the greater the acidity depletion with storage. This 2008) and most apple fruit consumed are subject to several phenomenon indicates physiological limitations of fruit months of storage, a breeding target of 3–4 Q alleles might acidity maintenance over long periods of storage, and has appear to be feasible. However, there are several arguments breeding implications. As organic acids are considered an against that strategy. First, the acidity associated with such energy source for fruit metabolism during storage (Etienne fruit is expected to be unpalatable at harvest. Second, the et al. 2013), the greater acidity depletion in fruit of culti- relatively rapid depletion of acidity in storage of these highest vars with higher Q-allele dosage might mean that high Q-allele dosages would be expected to be associated with an levels of titratable acidity are depleted at the expense of inconsistent perceived flavor over time, which could confuse other energy reserves. Third, the order of QTL genotypes consumers. Third, the rapid depletion of acidity would be for TA content remained the same during the examined expected to be associated with the shortest time within the storage period: no crossing regression lines. Therefore, fruit desired range, and thus the shortest market life. Indeed, acidity after storage appears to be a function of acidity at breeders appear to have avoided such a strategy as all nine harvest, which itself could be explained as a function of Q- important breeding parents in the present study carried only allele dosage. one or two Q alleles. And, from our observations over wider The apparent baseline of fruit acidity reached after extend- germplasm, few if any cultivars and selections, from heritage ed storage of fruit that was indicated by extrapolating beyond to modern, appear to have more than two Q alleles. Besides the experimentally observed results for each compound geno- too many Q alleles, zero Q alleles with its associated bland type has important implications, if true. If indeed TA de- flavor also appear to have been completely avoided in apple creases linearly over time in storage, all genotypes were pre- cultivar development. Because of the high frequency and mul- dicted to eventually reach the same level after about 9– tiple sources of Q and q alleles for both QTLs, there are nu- 12 months, with QTL genotype no longer influencing fruit merous ways that breeders are able to achieve the desired acidity. Fluctuations over the 3 years in the rate of loss and number of Q alleles in new cultivar development. Fine- storage time to reach the baseline might be due to particular tuning of the Q-allele dosage model to account for some epis- pre-harvest conditions experienced each year by the trees, as tasis between the two loci and effects of alternative (Q2) alleles they were all grown at the same location. These observations could further improve the explanatory power of the Q-allele implythatTAmeasuredatharvestforasetofcultivars dosage model. representing several Q-allele dosages could be used to predict acidity loss patterns throughout the subsequent storage sea- son. Further study is warranted to investigate whether fruit acidity decreases linearly with time in storage, and, if so, if Data archiving statement the arising predictive relationship that takes advantage of new genetic information on Q-allele dosage holds up in evalua- Genotypic data (1344 filtered SNP markers) of the multi- tions of fruit stored for durations between 2 and 12 months family germplasm set investigated has been submitted to the and in germplasm grown at other locations under other Genome Database for Rosaceae (www.rosaceae.org). Tree Genetics & Genomes (2019) 15:18 Page 15 of 17 18

Acknowledgements Dr. Luis F. Osorio provided helpful guidance with Gardiner SE, Bassil N, Peace C (2012a) Genome-wide SNP detec- ASReml statistical analysis, and Lichun Cai and Stijn Vanderzande tion, validation, and development of an 8K SNP array for apple. helped curate raw SNP data. Jack Klipfel conducted laboratory analyses PLoS One 7:e31745 for Ma1-SNP1455. Some genetic data and technical expertise came from Chagné D, Krieger C, Rassam M, Sullivan M, Fraser J, André C, Pindo the FruitBreedomics project no 265582: BIntegrated approach for increas- M,TroggioM,GardinerSE,HenryRA,AllanAC,McGhieTK, ing breeding efficiency in fruit tree crops^ (www.fruitbreedomics.com), Laing WA (2012b) QTL and candidate gene mapping for polyphe- which was co-funded by the EU seventh Framework Programme. nolic composition in apple fruit. BMC Plant Biol 12:12 Chagné D, Vanderzande S, Kirk C, Profitt N, Weskett R, Gardiner SE, Funding information This work was funded by the USDA’s National Peace CP,Volz RK, Bassil NV (2019) Validation of SNP markers for Institute of Food and Agriculture (NIFA)–Specialty Crop Research fruit quality and disease resistance loci in apple (Malus × domestica Initiative projects BRosBREED: Enabling marker-assisted breeding in Borkh.) using the OpenArray platform. Hortic Res (in press) Rosaceae^ (2009-51181-05808) and BRosBREED: Combining disease Cliff MA, Stanich K, Lu R, Hampson CR (2016) Use of descriptive resistance with horticultural quality in new rosaceous cultivars^ (2014- analysis and preference mapping for early-stage assessment of new 51181-22378), and USDA NIFA Hatch projects 0211277 and 1014919. and established apples. J Sci Food Agric 96:2170–2183 Costa e Silva J, Borralho NMG, Potts BM (2004) Additive and non- Open Access This article is distributed under the terms of the Creative additive genetic parameters from clonally replicated and seedling Commons Attribution 4.0 International License (http:// progenies of Eucalyptus globulus. Theor Appl Genet 108:1113– creativecommons.org/licenses/by/4.0/), which permits unrestricted use, 1119 distribution, and reproduction in any medium, provided you give Costa F, van de Weg WE, Stella S, Dondini L, Pratesi D, Musacchi S, appropriate credit to the original author(s) and the source, provide a link Sansavin S (2008) Map position and functional allelic diversity of to the Creative Commons license, and indicate if changes were made. Md-Exp7, a new putative expansin gene associated with fruit soft- eninginapple(Malus × domestica Borkh.) and pear (Pyrus – Publisher’snote Springer Nature remains neutral with regard to jurisdic- communis). Tree Genet Genomes 4:575 586 tional claims in published maps and institutional affiliations. Crane MB, Lawrence WJC (1934) Genetical studies in cultivated apples. JGenetics265–296 Daccord N, Celton JM, Linsmith G, Becker C, Choisne N, Schijlen E, van de Geest H, Bianco L, Muranty H, Micheletti D, Di Pierro EA, References Gouzy J, Gaillard S, Velasco R, Durel CE, Rees DJG, Laurens F, Lespinasse Y, Quesneville H, Aubourg S, Weigel D, van de Weg E, Bai Y, L, Li M, Fazio Z, Cheng L, Xu K (2012) A natural Troggio M, Bucher E (2017) High-quality de novo assembly of the mutation-led truncation in one of the two aluminum-activated ma- apple genome and methylome dynamics of early fruit development. – late transporter-like genes at the Ma locus is associated with low Nat Genet 49:1099 1106 fruit acidity in apple. Mol Gen Genomics 287:663–678 Di Guardo M, Micheletti D, Bianco L, Koehorst-van Putten HJ, Longhi S, Costa F, Aranzana MJ, Velasco R, Arús P,Troggio M, van de Weg Baojiang L, Guiyong L, Shixi J (1995) Studies on the inheritance and EW (2015) ASSIsT: an automatic SNP scoring tool for in- and selection of sweetness and acidity of apples. Acta Hortic 403:233– outbreeding species. Bioinformatics 31:3873–3874 240 Di Pierro EA, Gianfranceschi L, Di Guardo M, Koehorst - van Putten Beach SA, Booth NO, Taylor OM (1905) Apples of New York (Vol II) HJJ, Kruisselbrink JW, Longhi S, Troggio M, Bianco L, Muranty H, State of New York – Department of Agriculture. J. B. Lyon Pagliarani G, Tartarini S, Letschka T, Lozano-Luis L, Garkava- Company, New York, pp 150–152 Gustavsson L, Micheletti D, Bink MCAM, Voorrips RE, Aziz E, Bianco L, Cestaro A, Sargent DJ, Banchi E, Derdak S, Di Guardo N, Velasco R, Laurens F, van de Weg E (2016) A highdensity, multi- Salvi S, Viola R, Gut I, Laurens F, Chagné D, Velasco R, van de Weg parental SNP genetic map on apple validates a new mapping ap- E, Troggio M (2014) Development and validation of a 20K Single proach for outcrossing species. Hortic Res 3:16057 Nucleotide Polymorphism (SNP) whole genome genotyping array Dove WF, Murphy E (1936) The vitamin C-content of apples and its for apple (Malus × domestica Borkh). PLoS ONE 9: e110377 relation to human welfare. Science 83:325–327 Bink MCAM, Uimari P, Sillanpaa J, Janss LLG, Jansen RC (2002) Duan N, Bai Y, Sun H, Wang N, Ma Y, Li M, Wang X, Jiao C, Legall N, Multiple QTL mapping in related plant populations via a pedigree- Mao L, Wan S, Wang K, He T, Feng S, Zhang Z, Mao Z, Shen X, – analysis approach. Theor Appl Genet 104:751 762 Chen X, Jiang Y, Wu S, Yin C, Ge S, Yang L, Jiang S, Xu H, Liu J, Bink MCAM, Totir LR, ter Braak CJF, Winkler CR, Boer MP, Smith OS Wang D, Qu C, Wang Y, Zuo W, Xiang L, Liu C, Zhang D, Gao Y, (2012) QTL linkage analysis of connected populations using ances- Xu Y, Xu K, Chao T, Fazio G, Shu H, Zhong G-Y, Cheng L, Fei Z, tral marker and pedigree information. Theor Appl Genet 124:1097– Chen X (2017) Genome re-sequencing reveals the history of apple 1113 and supports a two-stage model for fruit enlargement. Nature Bink MCAM, Jansen J, Madduri M, Voorrips RE, Durel C-E, Kouassi Communications 8(1):249. https://doi.org/10.1038/s41467-017- AB, Laurens F, Mathis F, Gessler C, Gobbin D, Rezzonico F, 00336-7 Patocchi A, Kellerhals M, Boudichevskaia A, Dunemann F, Peil Etienne A, Génard M, Lobit P, Mbeguié-A-Mbéguié D, Bugaud C (2013) A, Nowicka A, Lata B, Stankiewicz-Kosyl M, Jeziorek K, Pitera What controls fleshy fruit acidity? A review of malate and citrate E, Soska A, Tomala K, Evans KM, Fernández-Fernández F, Guerra accumulation in fruit cells. J Exp Bot 64:1451–1469 W, Korbin M, Keller S, Lewandowski M, Plocharski W, Rutkowski Evans K (2013) Apple breeding in the Pacific Northwest. Acta Hort 976: K, Zurawicz E, Costa F, Sansavini S, Tartarini S, Komjanc M, Mott 75–78 D, Antofie A, Lateur M, Rondia A, Gianfranceschi L, van de Weg Evans K, Guan Y, Luby J, Clark M, Schmitz C, Brown S, Orcheski B, WE (2014) Bayesian QTL analyses using pedigreed families of an Peace C, van de Weg E, Iezzoni A (2012) Large-scale standardized outcrossing species, with application to fruit firmness in apple. phenotyping of apple in RosBREED. Acta Hortic 945:233–238 Theor Appl Genet 127:1073–1090 Gao Z, van de Weg WE (2006) The Vf gene for scab resistance in apple is Chagné D, Crowhurst RN, Troggio M, Davey MW, Gilmore B, Lawley linked to sub-lethal genes. Euphytica 151:123–132 C, Vanderzande S, Hellens RP, Kumar S, Cestaro A, Velasco R, Harker FR, Lau K, Gunson FA (2003) Juiciness of fresh fruit: a time- Main D, Rees J, Iezzoni A, Mockler T, Wilhelm L, van de Weg E, intensity study. Postharvest Biol and Technol 29:55–60 18 Page 16 of 17 Tree Genetics & Genomes (2019) 15:18

Harker FF, Kupferman EM, Marin AB, Gunson FA, Triggs FM (2008) (2009) Estimation of genetic parameters and prediction of breeding Eating quality standards for apples based on consumer preferences. values for apple (Malus x domestica) fruit quality traits using pedi- Postharvest Biol Technol 50:70–78 greed plant material in Europe. Tree Gen Genomes 5:659–672 Haynes D (1925) Chemical studies in the physiology of apples I. Change Kumar S, Chagne D, Bink MC, Volz RK, Whitworth C, Carlisle C (2012) in the acid content of stored apples and its physiological signifi- Genomic selection for fruit quality traits in apple (Malus x cance. Ann Bot 1:77–96 domestica Borkh.). PLoS One 7:e36674 Howard NP, van de Weg E, Bedford DS, Peace C, Vanderzande S, Clark Laurens F, Durel CE, Patocchi A, Peil A, Salvi S, Tartarini S, Velasco R, MD, Teh SL, Cai L, Luby JJ (2017) Elucidation of the ‘Honeycrisp’ van de Weg E (2010) Review on apple genetics and breeding pedigree through haplotype analysis with a multi-family integrated programmes and presentation of a new European initiative to in- SNP linkage map and a large apple (Malus × domestica)pedigree- crease fruit breeding efficiency. J Fruit Sci 27:102–107 connected SNP data set. Hortic Res 4:17003 Laurens F, Aranzana MJ, Arus P, Bassi D, Bink M, Bonany J, Caprera A, Howard NP, van de Weg E, Tillman J, Tong CBS, Silverstein KAT, Luby Corelli-Grappadelli L, Costes E, Durel CE, Mauroux JB, Muranty J (2018) Two QTL characterized for soft scald and soggy break- H, Nazzicari N, Pascal T, Patocchi A, Peil A, Quilot-Turion B, down in apple (Malus × domestica) through pedigree-based analysis Rossini L, Stella A, Troggio M, Velasco R, van de Weg E (2018) of a large population of interconnected families. Tree Genet An integrated approach for increasing breeding efficiency in apple Genomes 14:2 and peach in Europe. Hortic Res 5:11 Hu DG, Sun CH, Ma QJ, You CX, Cheng L, Hao YJ (2016) MdMYB1 Liebhard R, Gianfranceschi L, Koller B, Ryder CD, Tarchini R, van de regulates anthocyanin and malate accumulation by directly facilitat- Weg WE, Gessler (2002) Development and characterisation of 140 – ing their transport into vacuoles in apples. Plant Physiol 170:1315 new microsatellites in apple (Malus x domestica Borkh). Mol Breed 1330 10:217–241 Hu DG, Li YY, Zhang QY, Li M, Sun CH, Yu JQ, Hao YJ (2017) R2R3- Liebhard R, Koller B, Gianfranceschi L, Gessler C (2003) Creating a MYB transcription factor MdMYB73 is involved in malate accumu- saturated reference map for the apple (Malus × domestica Borkh.) – lation and vacuolar acidification in apple. Plant J 91:443 454 genome. Theor Appl Genet 106:1497–1508 Iezzoni A, Weebadde C, Luby J, Chengyan Y, van de Weg E, Fazio G, Ma B, Chen J, Zheng H, Fang T, Ogutu C, Li S, Han Y, Wu B (2015a) Main D, Peace C, Bassil N, McFerson J (2010) RosBREED: en- Comparative assessment of sugar and malic acid composition in – abling marker-assisted breeding in Rosaceae. Acta Hort 859:389 cultivated and wild apples. Food Chem 172:86–91 394 Ma B, Liao L, Zheng H, Chen J, Wu B, Ogutu C, Li S, Korban SS, Han Y Iwanami H, Moriya S, Kotoda N, Takahashi S, Abe K (2008) Storability (2015b) Genes encoding aluminum-activated malate transporter II in cold temperatures can be evaluated based on changes in fruit and their association with fruit acidity in apple. Plant Genome 8:0 quality in apple genotypes under shelf life conditions. HortScience Ma BQ, Zhao S, Wu BH, Wang D, Peng Q, Owiti A, Fang T, Liao L, 43:655–660 Ogutu C, Korban SS, Li S, Han Y (2016) Construction of a high Iwanami H, Moriya S, Kotoda N, Mimida N, Takahashi-Sumiyoshi S, density linkage map and its application in the identification of QTLs Abe K (2012) Mode of inheritance in fruit acidity in apple analysed for soluble sugar and organic acid components in apple. Tree Genet with a mixed model of a major gene and polygenes using large Genomes 12:1 complex pedigree. Plant Breed 131:322–328 Maliepaard C, Alston FH, van Arkel G, Brown LM, Chevreau E, Jia D, Shen F, Wang Y, Wu T, Xu X, Zhang X, Han Z (2018) Apple fruit Dunemann F, Evans KM, Gardiner S, Guilford P, van Heusden acidity is genetically diversified by natural variations in three hier- AW, Janse J, Laurens F, Lynn JR, Manganaris AG, den Nijs APM, archical epistatic genes: MdSAUR37, MdPP2CH and MdALMTII. Periam N, Rikkerink E, Roche P, Ryder C, Sansavini S, Schmidt H, Plant J 95:427–443 Tartarini S, Verhaegh JJ, Vrielink-van Ginkel M, King GJ (1998) JungS,FicklinSP,LeeT,ChengC-H,BlendaA,ZhengP,YuJ, Aligning male and female linkage maps of apple (Malus pumila Bombarely A, Cho I, Ru S, Evans K, Peace C, Abbott AG, Mill.) using multi-allelic markers. Theor Appl Genet 97:60–73 Mueller LA, Olmstead MA, Main D (2014) The Genome Database for Rosaceae (GDR): year 10 update. Nucleic Acids Res Mangandi J, Verma S, Osorio L, Peres NA, van de Weg E, Whitaker VM 42:D1237–D1244 (2017) Pedigree-based analysis in a multiparental population of oc- Kass RE, Raftery AE (1995) Bayes factors. J Am Stat Assoc 90(430): toploid strawberry reveals QTL alleles conferring resistance to – 773–795 Phytophthora cactorum. G3: Genes Genom Genet 7:1707 1719 Kenis K, Keulemans J, Davey M (2008) Identification and stability of Martinoia E, Maeshima M, Neuhaus HE (2007) Vacuolar transporters and – QTLs for fruit quality traits in apple. Tree Genet Genomes 4:647– their essential role in plant metabolism. J Exp Bot 58:83 102 661 Nybom N (1959) On the inheritance of acidity in cultivated apples. – Khan SA, Beekwilder J, Schaart JG, Mumm R, Soriano JM, Jacobsen E, Hereditas 45:332 350 Schouten HJ (2013) Differences in acidity of apples are probably Peace CP, Luby J, van de Weg E, Bink M, Iezzoni AF (2014) A strategy mainly caused by a malic acid transporter gene on LG16. Tree Genet for developing representative germplasm sets for systematic QTL Genomes 9:475–487 validation, demonstrated for apple, peach and sweet cherry. Tree King GJ, Maliepaard C, Lynn JR, Alston FH, Durel C-E, Evans KM, Genet Genomes 10:1679–1694 Griffon B, Laurens F, Manganaris AG, Schrevens E, Tartarini S, Pikunova A, Madduri M, Sedov E, Noordijk Y, Peil A, Troggio M, Bus Verhaegh J (2000) Quantitative genetic analysis and comparison of VGM, van de Weg E (2013) ‘Schmidt’s ’ is identical to physical and sensory descriptors relating to fruit flesh firmness in ‘Common Antonovka’, an apple cultivar widely used in Russia in apple (Malus pumila Mill.). Theor Appl Genet 100:1074–1084 breeding for biotic and abiotic stresses. Tree Genet Genomes 10: King GJ, Lynn JR, Dover CJ, Evans KM, Seymour GB (2001) 261–271 Resolution of quantitative trait loci for mechanical measures ac- Rouchaud L, Moons C, Meyer JA (1985) Cultivar differences in the counting for genetic variation in fruit texture of apple (Malus pumila influence of harvest date and cold storage on the free sugar and acids Mill.). Theor Appl Genet 102:1227–1235 contents, and the eating quality of apples. J Hortic Sci Biotech 60: Kouassi AB, Durel CE, Fabrizio C, Tartarini S, Van de Weg E, Evans K, 29l–296 Fernandez-Fernandez F, Govan C, Boudichevskaja A, Dunemann F, Silfverberg-Dilworth E, Matasci CL, van de Weg WE, Van Kaauwen Antofie A, Lateur M, Stankiewicz-kosyl M, Soska A, Tomala K, MPW, Walser M, Kodde LP, Soglio V, Gianfranceschi L, Durel Lewandowski M, Rutkovski K, Zurawicz E, Walter G, Laurens F CE, Costa F, Yamamoto T, Koller B, Gessler C, Patochhi A Tree Genetics & Genomes (2019) 15:18 Page 17 of 17 18

(2006) Microsatellite markers spanning the apple (Malus x Allan AC, Bus V, Chagné D, Crowhurst RN, Gleave AP, Lavezzo domestica Borkh.) genome. Tree Genet Genomes 2:202–224 E, Fawcett JA, Proost S, Rouzé P, Sterck L, Toppo S, Lazzari B, Sun R, Chang YS, Yang FQ, Wang Y, Li H, Zhao YB, Chen DM, Wu T, Hellens RP, Durel CE, Gutin A, Bumgarner RE, Gardiner SE, Zhang XZ, Han ZH (2015) A dense SNP genetic map constructed Skolnick M, Egholm M, van de Peer Y, Salamini F, Viola R using restriction site-associated DNA sequencing enables detection (2010) The genome of the domesticated apple (Malus × domestica of QTLs controlling apple fruit quality. BMC Genomics 16:747 Borkh.). Nat Genet 42:833–839 Sweetman C, Deluc LG, Cramer GR, Ford CM, Soole KL (2009) Visser T, Verhaegh JJ (1978) Inheritance and selection of some fruit Regulation of malate metabolism in grape berry and other develop- characters of apple. I. Inheritance of low and high acidity. ing fruits. Phytochemistry 70:1329–1344 Euphytica 27:753–760 van de Weg WE, Voorrips RE, Finkers HJ, Kodde LP, Jansen J, Bink Voorrips RE (2002) MapChart: software for the graphical presentation of MCAM (2004) Pedigree genotyping: a new pedigree-based ap- linkage maps and QTLs. J Hered 93:77–78 proach of QTL identification and allele mining. Acta Hortic 663: Voorrips RE, Bink MCAM, Kruisselbrink JW, Koehorst-van Putten HJJ, – 45 50 van de Weg WE (2016) PediHaplotyper: software for consistent van de Weg E, Di Guardo M, Jänsch N, Socquet-Juglard D, Costa F, assignment of marker haplotypes in pedigrees. Mol Breed 36:119 Baumgartner I, Broggini GAL, Kellerhals M, Troggio M, Laurens Watkins CB (2017) Advances in post-harvest storage and handling of F, Durel CE, Patocchi A (2018) Epistatic fire blight resistance QTL ‘ ’ apples. In: Evans K (ed) Achieving sustainable cultivation of apples. alleles in the apple cultivar Enterprise and selection X-6398 dis- Burleigh Dodds, Cambridge, pp 337–370 covered and characterized through pedigree-informed analysis. Mol Wellington R (1924) An experiment in breeding apples. II. - New York Breed 38:5 State Agr. Exp. Sta. Techn. Bull. No. 106. 149 pp Velasco R, Zharkikh A, Affourtit J, Dhingra A, Cestaro A, Kalyanaraman Xu K, Wang A, Brown S (2012) Genetic characterization of the Ma locus A, Fontana P, Bhatnagar SK, Troggio M, Pruss D, Salvi S, Pindo M, – Baldi P, Castelletti S, Cavaiuolo M, Coppola G, Costa F, Cova V, with pH and titratable acidity in apple. Mol Breed 30:899 912 Dal Ri A, Goremykin V, Komjanc M, Longhi S, Magnago P, Zhang YZ, Li PM, Cheng LL (2010) Developmental changes of carbo- Malacarne G, Malnoy M, Micheletti D, Moretto M, Perazzolli M, hydrates, organic acids, amino acids, and phenolic compounds in ‘ ’ – Si-Ammour A, Vezzulli S, Zini E, Eldredge G, Fitzgerald LM, Gutin Honeycrisp apple flesh. Food Chem 123:1013 1018 N, Lanchbury J, Macalma T, Mitchell JT, Reid J, Wardell B, Kodira Zhang Q, Ma BQ, Li H, Chang Y,Han Y,Li J, Wei G, Zhao S, Khan MA, C, Chen Z, Desany B, Niazi F, Palmer M, Koepke T, Jiwan D, Zhou Y, Gu C, Zhang X, Han Z, Korban SS, Li S, Han Y (2012) Schaeffer S, Krishnan V, Wu C, Chu VT, King ST, Vick J, Tao Q, Identification, characterization, and utilization of genome-wide sim- Mraz A, Stormo A, Stormo K, Bogden R, Ederle D, Stella A, ple sequence repeats to identify a QTL for acidity in apple. BMC Vecchietti A, Kater MM, Masiero S, Lasserre P, Lespinasse Y, Genomics 13:537